Dev Journal

How We Reduced AI Search Costs Through Query Caching

We implemented query caching, which significantly lowered our AI search expenses and improved system performance.

How We Searched for a Senior Developer and Explained Their Seniority

We faced challenges in finding a Senior Developer and explaining their level to recruiters, which led to changes in our hiring strategy.

How We Made Candidates Private by Default

We changed candidate privacy settings to ensure greater protection of personal information.

How We Rethought Our Approach to Calculating Match Score

We optimized the match score calculation algorithm, enhancing the quality of recommendations for users.

Why We Don't Send the Entire Candidate Database to GPT

We decided not to send the entire candidate database to GPT to avoid issues with privacy and the quality of recommendations.

How We Taught AI to Distinguish a Real Senior Go Developer from AI Hallucinations

We implemented deterministic filters and qualification signals to reduce AI errors in candidate evaluation.

Development note: synthetic example

This article documents an internal experiment related to a small product decision at Fitlane AI.